Neuroblastoma is a pediatric cancer with significant clinical heterogeneity. Despite extensive efforts, it is still difficult to cure children with high-risk neuroblastoma. Immunotherapy is a promising approach to treat children with this devastating disease. We have previously reported that macrophages are important effector cells in high-risk neuroblastoma. In this perspective article, we discuss the potential function of the macrophage inhibitory receptor SIRPA in the homeostasis of tumor-associated macrophages in high-risk neuroblastoma. The ligand of SIRPA is CD47, known as a “don’t eat me” signal, which is highly expressed on cancer cells compared to normal cells. CD47 is expressed on both tumor and stroma cells, whereas SIRPA expression is restricted to macrophages in high-risk neuroblastoma tissues. Notably, high SIRPA expression is associated with better disease outcome. According to the current paradigm, the interaction between CD47 on tumor cells and SIRPA on macrophages leads to the inhibition of tumor phagocytosis. However, data from recent clinical trials have called into question the use of anti-CD47 antibodies for the treatment of adult and pediatric cancers. The restricted expression of SIRPA on macrophages in many tissues argues for targeting SIRPA on macrophages rather than CD47 in CD47/SIRPA blockade therapy. Based on the data available to date, we propose that disruption of the CD47-SIRPA interaction by anti-CD47 antibody would shift the macrophage polarization status from M1 to M2, which is inferred from the 1998 study by Timms et al. In contrast, the anti-SIRPA F(ab’)2 lacking Fc binds to SIRPA on the macrophage, mimics the CD47-SIRPA interaction, and thus maintains M1 polarization. Anti-SIRPA F(ab’)2 also prevents the binding of CD47 to SIRPA, thereby blocking the “don’t eat me” signal. The addition of tumor-opsonizing and macrophage-activating antibodies is expected to enhance active tumor phagocytosis.